Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 12205
    Prev
    Next
    INDEX
    Explanations

    terms related to numerical data and statistics

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    525
    +0.13
    0.6%
    559
    +0.11
    0.5%
    56
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.13
    0.01
    284
    +0.11
    0.01
    525
    +0.10
    0.01
    Negative Logits
     emergency
    -0.60
    Spoiler
    -0.58
     spoilers
    -0.57
    assetsadobe
    -0.57
    vp
    -0.57
     surrogate
    -0.56
     homeland
    -0.56
     writing
    -0.56
     Uriel
    -0.55
     specificity
    -0.55
    POSITIVE LOGITS
    lyn
    1.01
    deals
    0.82
    merce
    0.81
    swick
    0.78
    amia
    0.78
    reau
    0.75
    angles
    0.71
    heet
    0.71
     trout
    0.71
    velt
    0.71
    Activations Density 0.024%

    No Known Activations